1. Field of the Invention
The present invention relates to a digital radio communication system, a reception system for a PDMA radio base station, and a method of calculating a weight vector. More particularly, the present invention relates to a digital radio communication system of the PDMA (Path Division Multiple Access) system having the function to calculate a weight vector by which a reception signal is multiplied to separate a desired signal from the signal received at a base station from a mobile station, reception system thereof, and weight vector calculation method.
2. Description of the Background Art
In the field of mobile communication systems such as portable telephones that have become extremely popular recently, various transmission channel allocation methods have been proposed to effectively use the frequencies. Some thereof are actually in practice.
FIGS. 6A-6C show the channel arrangement in various communication systems of Frequency Division Multiple Access (FDMA), Time Division Multiple Access (TDMA), and PDMA. The systems of FDMA, TDMA, and PDMA will be described briefly with reference to FIGS. 6A-6C.
Referring to FIG. 6A corresponding to the FDMA system, the analog signals of users 1-4 are frequency-divided to be transmitted in radio waves of different frequencies f1-f4. The signals of respective users 1-4 are separated by frequency filters.
Referring to FIG. 6B corresponding to the TDMA system, the digitized signals of respective users are time-divided and transmitted in radio waves of different frequencies f1-f4 at every constant period of time (time slot). The signals of respective users 1-4 are separated by frequency filters and by time synchronization between a base station and each user""s mobile terminal device.
Recently, the PDMA system has been proposed to improve the radio wave frequency usability to comply with the spread of portable telephones. In the PDMA system shown in FIG. 6C, one time slot of the same frequency is divided spatially to transmit data of a plurality of users. The signals of respective users 1-4 are separated using frequency filters, time synchronization between a base station and each user""s mobile terminal device, and a mutual interference removal apparatus such as adaptive arrays.
FIG. 7 shows the reception system of a conventional base station for use in PDMA. Four antennas 3-6 are provided to distinguish between user a and user b. The outputs of respective antennas are applied to frequency conversion circuits 7-10 to be frequency-converted by a local oscillation signal Lo and then applied to an A/D converter 11. The signals are converted into digital signals to be applied to a DSP (Digital Signal Processor) 12.
DSP 12 includes a channel allocation standard calculator 121, a channel allocation apparatus 122, and adaptive arrays 131 and 132. Channel allocation standard calculator 121 calculates data in advance to identify whether the signals of two users a and b can be separated by the adaptive arrays. In response to the calculation result, channel allocation apparatus 122 provides to each of adaptive arrays 131 and 132 the channel allocation information including user information that selects the frequency and the time. Each of adaptive arrays 131 and 132 is formed of, for example, a signal combine circuit as shown in FIG. 8. The signal of each user can be separated by selecting only the signal of a particular user.
FIG. 8 is a block diagram of a conventional adaptive array. In this example, in order to extract the signal of a certain user from input signals of a plurality of users, four input ports 14-17 are provided. The signals applied to input ports 14-17 are provided to a weight vector calculator 18 as well as to respective multipliers 20-23.
Weight vector calculator 18 uses the input signal as well as a training signal corresponding to the signal of a particular user prestored in a memory 19 or the output of an adder 24 that will be described afterwards to calculate weight vectors W1-W4. Multipliers 20-23 multiply the input signals of input ports 14-17 by weight vectors W1-W4, respectively. The multiplied results are sent to adder 24. Adder 24 adds the output signals of multipliers 20-23. The added result is output to an output port 25 and also to weight vector calculator 18 if necessary.
The weight vector will be described hereinafter.
Two signals X1 (t) and X2 (t) from a particular user are received at input ports 14 and 15 of the adaptive array of FIG. 8. Assuming that the adaptive array operates in an ideal manner, the output signal of the adaptive array is represented by the following equation.
Y (t)=W1X1 (t)+W2X2 (t)=S1 (t)+n (t)
Here, the weight vector W of that user is represented by the following equation.   W  =      [                                        W            1                                                            W            2                                ]  
FIGS. 9 and 10 schematically show the method of calculating a weight vector according to weight vector calculator 18 of the adaptive array of FIG. 8. FIG. 11 is a flow chart showing the procedure of signal extraction by the adaptive array of FIG. 8. As shown in FIG. 9, one time slot includes a preamble and a channel control signal transmitted through a control channel of a frequency fc, and a preamble and-data transmitted through a conversation channel of a frequency fT.
The weight vector of a user signal is calculated using the preamble and the data of the data signal transmitted via the conversation channel of FIG. 9.
The procedure of extracting a signal from a desired user using a weight vector will be described with reference to FIG. 11.
When extraction of a signal Ua (t) of user a applied to, for example, input port 14 is requested at step SP1 of FIG. 11, weight vector calculator 18 of FIG. 8 sets the initial values P (0) and W (0) of a correlation matrix P and a weight vector W as in the following equation at step SP2.             P      ⁢              xe2x80x83            ⁢              (        0        )              =                  δ                  -          1                    ⁡              [                                                            1                ,                0                ,                0                ,                0                                                                                        0                ,                1                ,                0                ,                0                                                                                        0                ,                0                ,                1                ,                0                                                                                        0                ,                0                ,                0                ,                1                                                    ]              ,      xe2x80x83    ⁢            W      ⁢              xe2x80x83            ⁢              (        0        )              =          [                                    1                                                0                                                0                                                0                              ]      
Here, xcex4 is a positive decimal, for example xcex4=1.0xe2x88x9210.
At step SP3, time t=1 is set. At step SP4, the Kalman gain vector K (t) at time t is calculated according to the following equation.
T(t)=xcexxe2x88x921P(txe2x88x921 )X(t)
      K    ⁢          xe2x80x83        ⁢          (      t      )        =            T      ⁢              xe2x80x83            ⁢              (        t        )                    1      +                        X          H                ⁢                  xe2x80x83                ⁢                  (          t          )                ⁢                  xe2x80x83                ⁢        T        ⁢                  xe2x80x83                ⁢                  (          t          )                    
In the above equations, T (t) indicates the intermediate generated vector at time t.
At step SP5, determination is made whether the time length Tp of the preamble shown in FIG. 9 is smaller than t or not. When txe2x89xa7Tp, a standard signal d (t) is calculated according to the following equation at step SP6.
d(t)=det[W(txe2x88x921) TX (t)]
X (t) indicates the reception signal vector at time t.
When not txe2x89xa7Tp at step SP5, control proceeds to step SP7 to substitute a training signal dTr (t) for standard signal d (t). At step SP8, error e (t) of time t is calculated according to the following equation.       e    ⁢          xe2x80x83        ⁢          (      t      )        =            d      ⁢              xe2x80x83            ⁢              (        t        )              -                  [                                            W              1                        ⁢                          xe2x80x83                        ⁢                          (              t              )                                ,                                    W              2                        ⁢                          xe2x80x83                        ⁢                          (              t              )                                ,                                    W              3                        ⁢                          xe2x80x83                        ⁢                          (              t              )                                ,                                    W              4                        ⁢                          xe2x80x83                        ⁢                          (              t              )                                      ]            ⁢              xe2x80x83            [                                                                  X                1                            ⁢                              xe2x80x83                            ⁢                              (                t                )                                                                                                        X                2                            ⁢                              xe2x80x83                            ⁢                              (                t                )                                                                                                        X                3                            ⁢                              xe2x80x83                            ⁢                              (                t                )                                                                                                        X                4                            ⁢                              xe2x80x83                            ⁢                              (                t                )                                                        ]      
At step SP9, weight vector W (t) at time t is calculated according to the following equation.
W(t)=W (t)+e(t) K (t)
At step SP10, correlation matrix P (t) at time t is calculated according to the following equation.
P(t)=xcexxe2x88x921P (t)xe2x88x92K(t) TH (t) 
At step SP11, signal Ua (t) of user a at time t is calculated according to the following equation.       Ua    ⁢          xe2x80x83        ⁢          (      t      )        =            [                                    W            1                    ⁢                      xe2x80x83                    ⁢                      (            t            )                          ,                              W            2                    ⁢                      xe2x80x83                    ⁢                      (            t            )                          ,                              W            3                    ⁢                      xe2x80x83                    ⁢                      (            t            )                          ,                              W            4                    ⁢                      xe2x80x83                    ⁢                      (            t            )                              ]        ⁢          xe2x80x83        [                                                      X              1                        ⁢                          xe2x80x83                        ⁢                          (              t              )                                                                                      X              2                        ⁢                          xe2x80x83                        ⁢                          (              t              )                                                                                      X              3                        ⁢                          xe2x80x83                        ⁢                          (              t              )                                                                                      X              4                        ⁢                          xe2x80x83                        ⁢                          (              t              )                                            ]  
At step SP12, determination is made whether the time length T of one time slot is txe2x89xa7T. When txe2x89xa7T is not established, control proceeds to step SP13 to increment t by +1. Then, control returns to step SP4.
In the conventional weight vector calculation method of FIG. 11, calculation of the weight vector must be converged within time period Tp of the preamble of the data signal included in the conversation channel of FIG. 9. However, calculation of the weight vector will not converge within time period Tp if time period Tp of the preamble is short. As a result, the reception property is degraded.
Furthermore, in the conventional calculation method of a weight vector shown in FIG. 10, the weight vector of the received data signal was calculated using the same algorithm throughout one time slot. There is a disadvantage that the amount of calculation is great when a RLS (Recursive Least Square) algorithm, for example, is used. When the CMA (Constant Modulus Algorithm) is used, the convergence of the weight vector is time consuming. A long time is required until proper data is output, so that an error may occur in the data.
In view of the foregoing, an object of the present invention is to prevent degradation in the property of the reception signal in a digital radio communication system by reducing the time for convergence of the weight vector calculation.
Another object of the present invention is to improve the communication quality with an appropriate amount of calculation by appropriately switching the algorithm during weight vector calculation.
According to an aspect of the present invention, a radio communication system transmitting a control signal to initiate conversation through a control channel and transmitting a data signal for conversation through a subsequent conversation channel having a frequency differing from that of the control channel, includes a reception circuit with a plurality of antennas to receive a series of signals constituted by the control signal transmitted through the control channel and the data signal transmitted through the conversation channel from a user, a first calculation circuit for calculating a weight vector of the control signal received through the control channel, a memory for storing the calculated weight vector of the control signal, a second calculation circuit for calculating a weight vector of the data signal received through the conversation channel with the weight vector stored in the memory as an initial value, and a signal extraction circuit for extracting a data signal according to the calculated weight vector of the data signal.
According to another aspect of the present invention, a reception system of a PDMA radio base station receiving from each user a series of signals constituted by a control signal to initiate conversation transmitted through a control channel and a data signal for conversation transmitted through a subsequent conversation channel having a frequency differing from that of the control channel, includes a plurality of antennas receiving the series of signals from each user, and an adaptive array separating a series of signals of a desired user from the signals received on the antennas. The adaptive array includes a first calculation circuit for calculating a weight vector of the control signal received through the control channel, a memory for storing the calculated weight vector of the control signal, a second calculation circuit for calculating a weight vector of the data signal received through the conversation channel with the weight vector stored in the memory as an initial value, and a signal extraction circuit for extracting the data signal according to the calculated data signal weight vector.
According to a further aspect of the present invention, a weight vector calculation method in a radio communication system that transmits a control signal to initiate conversation through a control channel and that transmits a data signal for conversation through a subsequent conversation channel having a frequency differing from that of the control channel, includes the steps of receiving a series of signals constituted by the control signal transmitted through the control channel and the data signal transmitted through the conversation channel from a user using a plurality of antennas, calculating a weight vector of the control signal received through the control channel, storing the calculated weight vector of the control signal, calculating a weight vector of the data signal received through the conversation channel with the stored weight vector as an initial value, and extracting the data signal according to the calculated weight vector of the data signal.
According to still another aspect of the present invention, a weight vector calculation method in a radio communication system that transmits a data signal formed of signal portions with different contents through a conversation channel within one time slot, includes the steps of receiving the data signal transmitted through the conversation channel from a user using a plurality of antennas, and calculating a weight vector using algorithms different between the signal portions having different contents.
The main advantage of the present invention is that the time required for the convergence of weight vector calculation can be reduced by calculating beforehand a weight vector in a control channel and using the vector value thereof as an initial value in calculating a weight vector in a conversation channel.
Another advantage of the present invention is that communication quality is improved with an appropriate amount of calculation by switching the algorithm to calculate a weight vector between the preamble and the data forming a data signal in the conversation channel.
The foregoing and other objects, features, aspects and advantages of the present invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings.